The Susceptibility of Example-Based Explainability Methods to Class Outliers
This work addresses robustness issues in explainable AI for practitioners, but it is incremental as it builds on existing metrics and methods.
The study investigated how class outliers affect example-based explainability methods for black-box ML models, finding that current methods, including those designed to suppress outliers, show significant shortcomings in experiments on text and image classification datasets.
This study explores the impact of class outliers on the effectiveness of example-based explainability methods for black-box machine learning models. We reformulate existing explainability evaluation metrics, such as correctness and relevance, specifically for example-based methods, and introduce a new metric, distinguishability. Using these metrics, we highlight the shortcomings of current example-based explainability methods, including those who attempt to suppress class outliers. We conduct experiments on two datasets, a text classification dataset and an image classification dataset, and evaluate the performance of four state-of-the-art explainability methods. Our findings underscore the need for robust techniques to tackle the challenges posed by class outliers.